The invention discloses a key quality characteristic decoupling analysis method considering risk criticality. The method comprises the following steps: constructing a coupling matrix of key quality characteristics by utilizing a modular design method; determining key quality characteristic risk criticality sorting through a propagation model, and identifying a key path of the propagation model; performing clustering sensitivity analysis on the coupling matrix by adopting a fuzzy clustering algorithm; taking the risk criticality after propagation as a basis for selecting an optimal clustering result; and carrying out module division and reconstruction on the coupling matrix so as to convert the coupling matrix into an approximate decoupling design matrix. According to the method, the risk criticality index of the key quality characteristics and the similarity index of the fuzzy clustering are used as the selection basis of the optimal clustering result, so that the decoupling analysis result is closer to the engineering practice and is more objective and accurate, and the subjectivity of selecting a plurality of clustering results is avoided. The method provides a theoretical basisfor reducing the quality risk caused by the coupling relationship.